Bilal’s Macroscope: Live with Models, Die with Models

Isaac Newton was a brilliant scientist with notable insights into gravity and optics. Less well-known is his foray into investing. Despite his scientific mind, he invested in the biggest bubble of the 18^th century – the South Sea bubble. But more than that, he tried to find a ‘model’ to convert lead into gold. At the time, this ‘model’ was called alchemy. In his notes, he thought he could use ‘sophick mercury’. Had it worked, he would have created a perpetual money-making machine. This quest continues today with investors using the latest in artificial intelligence (AI) to reach this elusive goal. Sadly, just like the imaginary perpetual motion machine, we think the nature of markets is such that the goal is impossible.

Models Behind Market Moves

Thursday’s market moves exemplify this challenge. Rates momentum models – one of the simplest forms of AI models – delivered 17% annual returns with a Sharpe ratio of 2.3 up until Wednesday. In FX, such models have performed extremely well in USD/JPY (+27%). Huge amounts of money are likely following versions of these models and all of them were positioned short US bonds and long USD/JPY. But Thursday’s US CPI data came out lower-than-expected, which triggered an initial rates rally. This in turn caused these models to unwind their positions, which induced further selling and a market rout. US 10-year yields fell over 28bps on Thursday – the third largest drop since 2009. Only the Federal Reserve’s (Fed) first QE programme (March 2009) and COVID (March 2020) saw larger drops. Meanwhile, USD/JPY saw its largest drop since 2009.

Why Models Fail but Survive

Herein lies the challenge of modelling financial markets even with the latest AI – any model that starts to trade markets becomes a part of the market. There is an endogeneity in markets that other fields, where AI is applied, do not have. For example, if an AI has to determine the difference between a picture, say of an apple or a banana, the AI does not interfere with the picture. It is an external observer. However, in financial markets any type of investor, whether AI or human, becomes part of the market. Therefore, one needs to model that effect, but then you need to model the model that is modelling the models’ effect, and so on. Keynes called this the beauty parade; George Soros called it market reflexivity and the typical investor would call it ‘what’s priced’ or ‘what’s the consensus trade’.

Nevertheless, machines will dominate over humans in trading thanks to massive advances in computer power, cloud services, trading APIs and AI modelling. And given most investors work with the same data, they will end up gravitating towards the same models that have worked recently. So, there will be herding around models. And when these models turn, you get outsized moves like Thursday. The challenge for any investor, then, is to determine the meaning of such market moves. Is it that most of the price was ‘technical’ in nature so does not reflect a change in fundamentals? Or was the price move fully reflective of fundamentals?

What Next

Put another way, at face value, the scale of Thursday’s moves suggests that the US inflation dragon has been slayed, the Fed is almost at the end of its hiking cycle and there will be no deep recession. Or the moves were reflective of previously successful trading models being crushed, forced position unwinds and panic. In the immediate aftermath, it is hard to know which is true; the truth likely lies somewhere in between. But if inflation history is any guide, slaying the inflation dragon is never that easy. At the same time, we need to estimate how much of the models’ positions have been unwound. For now, we are keeping our heads down and focusing on cross-market trades rather than outright rates or equity trades.

Our Current Discretionary Trades

We are in the process of reviewing our short EUR/USD and our rates trades, but exit our other long dollar trades (INR and CNH), otherwise we hold our cross-trades, which include:

1. US rates flattener: sell 2Y, buy 10Y
2. Short GBP/CHF: peak UK optimism
3. Short AUD/NZD: dovish RBA and weak China vs hawkish RBNZ
4. EM FX crosses: long THB/TWD and short ZAR/IDR, long SGD/CNH
5. Reviewing USD: short EUR/USD
6. Reviewing short rates: short bunds, paid PLN, CZK, THB, INR 1Y or 2Y rates

What Are Models Signalling?

We track a range of models which can help understand where the market is in pain:

• The best-performing momentum models, before Thursday’s moves, were short rates (+17%) and long dollar (+27% for USD/JPY). That is where we have also seen the biggest reversals. Equity momentum models have done less well, so there was likely less positioning there.
• Our FX carry models had delivered stellar returns of up to 30% over the past year. We would be wary of these trades now – and suggest that currencies like JPY, EUR, CHF could outperform high-yielders like USD in G10, and INR, MXN and HUF in EM.
• Our China Growth Tracker is showing signs of stabilising which likely reflects hopes of the end of zero-COVID.
• Our Flight Tracker is showing tentative signs of China flights bottoming, while Thailand flights have surged.

Recent Questions from Clients

• How much of hedge fund and CTA positions have been unwound?
• Has the dollar peaked?
• Are we close to de-escalation in the Russia-Ukraine war?
• Is China going to re-open soon? (A weekly question it seems!)
• Does the FTX fiasco have any contagion risks?
• What are your favourite dollar shorts?

What I Am Reading and Watching

• With all the talk of gender identity, I wrote a blog on identity and the metaverse.
• Social Media and Teenage Mental Health: Quasi-Experimental Evidence. This paper finds ‘internet significantly increased teen girls’ mental health [illness] relative to teen boys over the period when …social media became dominant’.
• Personality differences from day one after birth despite identical genes and identical environment. Fascinating. The paper suggests ‘individuality in later life can be strongly shaped by prenatal factors, such as the nourishment in the womb or other maternal effects, epigenetics and pre-birth developmental stochasticity’.
• Shark Lands on Fishing Boat. The boat people seem very relaxed.
• Demolition Photographer Uses Large Format Camera to Capture Explosive Images. Powerful pictures.
• ‘The Office’ Star Rainn Wilson Has Changed His Name To Raise Awareness For The Environmental Crisis. Dwight from the Office!
• You can now play pickleball at JFK airport. It is everywhere.
• I have started the TV show Your Honor – it is gripping. And I am still loving Gangs of London

Bilal Hafeez is the CEO and Editor of Macro Hive. He spent over twenty years doing research at big banks – JPMorgan, Deutsche Bank, and Nomura, where he had various “Global Head” roles and did FX, rates and cross-markets research.

(The commentary contained in the above article does not constitute an offer or a solicitation, or a recommendation to implement or liquidate an investment or to carry out any other transaction. It should not be used as a basis for any investment decision or other decision. Any investment decision should be based on appropriate professional advice specific to your needs.)